© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-27B-IT
    5. 56-GEMMASCOPE-2-TRANSCODER-262K
    7. 206672
    Prev
    Next
    INDEX
    Explanations

    key concepts

    np_max-act-logits · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_56_width_262k_l0_small_affine
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     red
    1.04
    red
    0.84
     reds
    0.80
    红
    0.79
    紅
    0.79
    レッド
    0.71
     레드
    0.68
    Red
    0.66
    RED
    0.65
     крас
    0.64
    POSITIVE LOGITS
     debates
    0.54
     Debate
    0.52
     debating
    0.52
    deb
    0.50
     debate
    0.49
     debated
    0.48
     Deb
    0.47
    辩
    0.46
    debate
    0.46
    Deb
    0.45
    Activations Density 0.008%

    No Known Activations